← Business portfolio
AI & integrations / Custom platform

LotPilot

Turn property documents into structured data.

PROJECT TYPE

AI & integrations

BUILT WITH
Next.jsSupabaseOpenAIMCP

The problem

Auction information arrives in PDFs, photographs, web pages and free text. Extracting it is only part of the task: a new document may describe an existing lot, corrections need review, and records must stay within the right organisation.

An AI response should not become a permanent property record simply because it looks plausible.

The solution

LotPilot turns source material into structured candidate records, validates their shape and presents a review table showing new lots and proposed updates. The operator chooses what to import before the database changes.

Imports match auctions and lot numbers so repeated processing can update the same record instead of creating duplicates. Source-document references preserve provenance, and database permissions separate organisations and distinguish editors from viewers.

The decisions behind it

Extraction before commitment
AI prepares a candidate result. Validation and a human review step sit between that result and the auction database.
Re-import means reconcile
The same document may be processed more than once. Stable matching rules decide whether a row creates a lot or updates one.
WHAT IT BRINGS TOGETHER
  • Auction, lot and apartment management
  • AI extraction with review before saving
  • Organisation-scoped MCP integration
Discuss a similar project